Package: trama.multi 0.3.8

trama.multi: Multivariate Analysis Collection for trama

Nodes for multivariate analysis in a trama graph: principal component analysis with scree, biplot and correlation circle; factor analysis by maximum likelihood or principal axis, with orthogonal (varimax, quartimax, equamax) and oblique (promax, oblimin) rotations; linear and quadratic discriminant analysis with exact leave-one-out predictions; logistic regression (binary and multinomial) as a classifier; jackknife bias, standard errors and pseudovalues for component, factor, discriminant and logistic statistics; and the diagnostics that come first (KMO, Bartlett's sphericity, Horn's parallel analysis, Box's M); distance matrices (Euclidean, Mahalanobis D2, Gower), hierarchical clustering with cophenetic correlation, k-means and Tocher's optimization method; and one-way MANOVA with optional blocks. Ships example data with known structure. Builds on trama.data for tables, trama.view for plots and trama.models, whose model contract the classifiers implement (prediction, confusion and ROC are its nodes).

Authors:Pedro Mambelli Fernandes [aut, cre], João Paulo Assis Bonifácio [aut]